Food Research International 49 (2012) 687–693
Contents lists available at SciVerse ScienceDirect
Food Research International journal homepage: www.elsevier.com/locate/foodres
The impact of freeze-drying on microstructure and rehydration properties of carrot Adrian Voda a,⁎, Natalia Homan b, Magdalena Witek b, 1, Arno Duijster c, Gerard van Dalen a, Ruud van der Sman d, Jaap Nijsse a, Lucas van Vliet c, Henk Van As b, John van Duynhoven a, b a
Unilever Research and Development Vlaardingen, Olivier van Noortlaan, P.O. Box 114, 3130 AC Vlaardingen, The Netherlands Laboratory of Biophysics and Wageningen NMR Centre, Wageningen University, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands Quantitative Imaging Group, Delft University of Technology, Lorentzweg 1, 2628 CJ Delft, The Netherlands d Food Process Engineering, Wageningen University, Bomenweg 2, 6703 HD Wageningen, The Netherlands b c
a r t i c l e
i n f o
Article history: Received 18 June 2012 Accepted 22 August 2012 Keywords: MRI NMR Microtomography μCT Microscopy SEM Image analysis Pore size distribution Tortuosity Microstructure Winter carrot Freeze-drying
a b s t r a c t The impact of freeze-drying, blanching and freezing rate pre-treatments on the microstructure and on the rehydration properties of winter carrots were studied by μCT, SEM, MRI and NMR techniques. The freezing rate determines the size of ice crystals being formed that leave pores upon drying. Their average size (determined by μCT) can be predicted in a quantitative manner by considering dendritic growth and freezing rates. Blanching as a pre-treatment, however, did not affect pore size distribution induced by freeze-drying. Upon rehydration of the freeze-dried carrots, PFG NMR and MRI show that cellular compartments were not restored and instead a porous network with permeable barriers is formed. Blanching pre-treatment introduced a less connected and more anisotropic porous network if followed by fast freezing, indicating that more of the native cell wall morphology is preserved. © 2012 Elsevier Ltd. All rights reserved.
1. Introduction Consumers have a high appreciation for fruits and vegetables, which are an important dietary source of vitamins, phytochemicals, fibers and minerals (Hoffmann, Boeing, Volatier, & Becker, 2003). The intake of fruits and vegetables has been associated with a wide range of beneficial health effects (Pomerleau, Lock, & Mckee, 2006). A main hurdle for consumers to raise their daily intake is the lack of convenience in preparing meals. The food industry has addressed this by offering the consumer dried fruits and vegetables, which are rehydrated shortly before consumption. A major obstacle for further growth in this area is the relative poor quality of the rehydrated fruits and vegetables in the product after preparation. Another bottleneck is the poor compromise between convenience in meal preparation and textural quality (Jangam, 2011; Prothon, Ahrne, & Sjoholm, 2003). Most dehydrated fruits and vegetables are produced by air drying. A disadvantage of this method is a substantial degradation in quality, including appearance (shrinkage, drying-up, darkening), nutrients, ⁎ Corresponding author. Tel.: +31 10 4606708; fax: +31 10 4606545. E-mail address:
[email protected] (A. Voda). 1 Present address: Institute of Physics, Jagiellonian University, Reymonta 4, 30‐059 Kraków, Poland. 0963-9969/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.foodres.2012.08.019
flavor, and the low rate of rehydration (Devahastin & Niamnuy, 2010; Ratti, 2001). Higher quality products can be obtained using more expensive freeze-drying methods (Mujumdar & Law, 2010). Freeze-drying involves crystallization of water in ice crystals, which subsequently sublimate, thus leaving a porous dried product. This may lead to loss in texture and an increase in friability (Brown, 1976; Chassagne-Berces et al., 2009; Ratti, 2001; Van Buggenhout et al., 2006). Improvements in the freeze-drying process of foods have been driven by engineering, where technologies are being optimized to balance rehydration rate and final texture (Mujumdar, 2011; Sagar & Kumar, 2010). Considering the underlying microstructure and its role in rehydration may enhance the efficiency and rate of process innovation (Mebatsion, 2008). A major barrier to embark on such an approach has been the lack of adequate quantitative measurement technologies that enable decision making based on sound microstructural data. Hence we embarked on an approach where we quantitatively assessed microstructural features of freeze-dried carrots as a model system. In this work the impact of thermal pre-treatments and freeze-drying on the microstructure of the cortical tissue of winter carrots was investigated. The purpose of this investigation was to quantitatively describe the features of dry and rehydrated microstructures by means of dedicated image analysis and NMR parameters. To achieve this, a suite
688
A. Voda et al. / Food Research International 49 (2012) 687–693
of imaging techniques in combination with NMR relaxometry and diffusometry tools was employed. In order to visualize the microstructure of the dried carrots at the μm–mm level X-ray computerized tomography (μCT) and scanning electron microscopy (SEM) were used. μCT allows for high-resolution 3D visualization and characterization of the dried and hydrated material (van Dalen, Notenboom, van Vliet, Voortman, & Esveld, 2007). μCT can probe the microstructure of samples non-invasively with an axial and lateral resolution down to a few micrometers and a field of view of up to a few mm under environmental conditions. Time-domain NMR and MRI were used to assess mobility of water in rehydrated samples in a non-invasive manner. Compartment integrity and tissue permeability in plant materials have been studied by 2D relaxometry and MRI. Time-domain NMR relaxometry (van Duynhoven, Voda, Witek, & As, 2010) has already been used to study freeze-drying of carrots (Hills & Nott, 1999). Relaxometry could detect sublimation of the frozen core and removal of non-frozen water during freeze-drying. A similar approach was used to study osmotic dehydration of apple (Cornillon, 2000). To obtain an indication if one has good control over the microstructure via process conditions of the freezing step during freeze-drying, the size of the pores of the freeze-dried samples, as obtained by image analysis, was compared with scaling rules for ice crystal size induced by dendritic growth. The process conditions were characterized by the heat transfer coefficient of the coolant and the freezing (coolant) temperature, which can be reformulated in terms of freezing rate. In our comparison we also used literature data of ice crystal size growth in comparable materials at different freezing rates.
2. Materials and methods 2.1. Materials The carrots used in the current study were of the winter carrot type purchased in a local supermarket, having sizes of 3 to 6 cm in diameter and a length of 20 to 30 cm. The carrot root consists of two distinct layers: a central stele and a peripheral cortex. The anatomy of the carrot root is greatly affected by the growth of the stele, hence four different regions can be distinguished as shown in Fig. 1: three inner regions belonging to the stele (vascular tissue) and the
outer region which is the cortex. Cylindrical samples with a diameter of 8 mm and a length of 10 mm were cut as it is shown in Fig. 1. The selected tissue consists of cells with a diameter of 20 to 100 μm, depending on the age of the carrot. 2.2. Pretreatments and freeze-drying of carrots Certain thermal pretreatments were applied to the carrot samples before freeze-drying. The first pretreatment was blanching for one minute in boiling water. Nonblanched samples were also selected for the next step. Secondly, samples were frozen at four different temperatures: − 28, − 80, − 150 and − 196 °C. The freeze-drying protocol consisted of time-incremented temperature steps from − 30 °C up to 25 °C at a constant low pressure of 0.4 mbar. The time needed to complete this process was about 27 h, see Table S1 in the supporting information for full details. 2.3. Scanning electron microscopy (SEM) of freeze-dried carrots A piece of dried carrot was cut into two halves in such a way that a cross-section was obtained. A very thin slice was cut off from the surface with a razor blade to obtain a high quality cross-sectional surface of the remaining piece of dry tissue. This surface was sputter coated with platinum for better SEM imaging quality. The Pt coated sample was inserted into a Jeol 6490 LA scanning electron microscope and both the peripheral and central areas were imaged at magnifications ranging from 10 × to 1000 ×. 2.4. X-ray microscopic computerized tomography (μCT) of freeze-dried carrots The internal porous structures of the samples were visualized using a SkyScan 1172 desktop X-ray micro-tomography system (Belgium, http:// www.skyscan.be). The X-ray tube was operated at 39 kV and 248 μA. μCT is a non-invasive method for the reconstruction of a three-dimensional volumetric image based on a large number of two-dimensional projections of a sample under different angles. The sample was imaged in a plastic cylindrical sample holder with an inner diameter of 11 mm. A stack of 2481 flat cross sections (4000×4000 pixels) was obtained after tomographic reconstruction of projection images (4000×2096 pixels). These projections were acquired under different rotations over a 180 degree interval with a step size of 0.20° (frame averaging=2). The acquisition time for one projection was 589 ms (exposure) resulting in a total acquisition and read-out time per scan of about 50 min (1019 projection images/scan). For tomographic reconstruction the following settings were used: no smoothing, ring artifact correction=20 and beam hardening correction=40%. A pixel size of 4.0 μm was selected. The samples were scanned using two scans connected in the vertical direction to increase the axial field-of-view (oversized scan) and subsequently merged together during reconstruction. 2.5. Nuclear magnetic resonance (NMR) of rehydrated carrots
Fig 1. Schematic representation of the four distinct tissue regions of a winter carrot: 1—parenchyma cell type, 2—parenchyma cells and xylem vessels, 3—outer cells of the vascular tissue and phloem vessels, and 4—cortex tissue. The disk indicates the sampling spot for the cylindrical samples employed in this work.
Carrot samples were rehydrated in water at 95 °C for 15 min. After rehydration, the samples were gently rolled on an absorbent paper tissue to remove the water dripping over the surface, and wrapped in cling-film to prevent moisture loss during measurements. Time-domain NMR relaxation experiments were carried out at room temperature (22 °C) on a Maran (Resonance Instruments, Oxford, UK) spectrometer operating at a proton frequency of 30 MHz. Transverse relaxation times were measured using a Carr– Purcell–Meiboom–Gill (CPMG) pulse sequence with echo-times (TE) of 400 μs and a repetition time (TR) of 5 s. Data were averaged over 64 acquisitions. T2 distributions were calculated as a continuous distribution of exponentials by means of the CONTIN software (Provencher, 1982), as well as a discrete analysis assuming a sum
A. Voda et al. / Food Research International 49 (2012) 687–693
of a limited number of exponentials by non-linear least square fitting was performed. Magnetic resonance imaging (MRI) and self-diffusion experiments were carried out at 7 T field strength (300 MHz proton operating frequency) with a Bruker Avance III spectrometer. The microimaging and self-diffusion hardware employed consists of a 10 mm birdcage type resonator and a field gradient unit allowing for a maximum gradient strength of 1.5 T/m. Rehydrated carrot samples were placed in NMR glass tubes and experiments were performed at room temperature. For assessment of water self-diffusion behavior, pulsed field gradient stimulated echo pulse sequence (PFG-StE) with bipolar gradients was used with gradient duration kept constant at 2 ms, while the gradient strength was varied between 0 and 1.5 T/m in 12 logarithmic distributed steps. The repetition time was set to 6 s and the signal was averaged over 16 scans. The self-diffusion observation time spanned an interval between 20 and 1000 ms. At short observation times t, the self-diffusion constant of water D(t) is determined by the surface-to-volume ratio of the pore space. At long times, the molecules probe the connectivity of the pore matrix and also the permeability of the pore walls, and D(t) is determined by a tortuosity parameter weighted by the permeability of the rehydrated cell wall material. The behavior of the self-diffusion coefficient in time as described in the literature (Sen, 2004; Sibgatullin, Vergeldt, Gerkema, & Van As, 2010; van der Weerd, Melnikov, Vergeldt, Novikov, & Van As, 2002) can be exploited to estimate the tortuosity of the pore network. Self-diffusion was also measured in both longitudinal and transversal directions, thus allowing for the assessment of network anisotropy. Longitudinal and transversal spin density images were recorded by means of a 2D-RARE pulse sequence (Bernstein, King, & X. J. Z., 2004) with a selected slice thickness of 500 μm. The in-plane field of view (FOV) was 12 × 12 mm 2. The complete k-space was covered in 128 phase encoding steps and 128 read points, resulting in a spatial resolution of 78 μm/pixel. The slice selection has been achieved with Gaussian shaped radio frequency (r.f.) excitation/refocusing pulses of 1 ms / 0.58 ms duration. The echo-time (TE) was set to 3 ms; the acquisition bandwidth to 50 KHz, and the repetition time to 2 s with crusher gradients following each acquisition to destroy residual transverse magnetization from components with long relaxation times. 2.6. Image analysis of μCT images of freeze-dried carrots The gray-valued μCT images were thresholded in order to get a binary image. After thresholding, the images were despeckled and masked in order to define the outer edge of the carrot sample. The large voids and the cracks extending toward the edge are regarded to be part of the sample. The acquired binary images were used as an input for the calculation of the pore size distribution. The granulometry calculations were performed on the three-dimensional datasets (4000×4000×2000 voxels) using morphological sieving (Bangham, Chardaire, Pye, & Ling, 1996; Luengo Hendriks, van Kempen, & van Vliet, 2007). Pore size distributions were acquired by a closing scale-space, using a sphere with increasing diameter. The spherical shape of the structuring element guarantees rotation invariance, so it is independent of the orientations of the sample and of the structures within the sample. The morphological closing is now defined as a dilation of the image with a specific structuring element, subsequently followed by erosion with the same structuring element C ðx; dÞ ¼ ðI ðxÞ⊕SðdÞÞ⊖SðdÞ;
ð1Þ
where ⊕ and ⊖ denote the dilation and the erosion respectively, I(x) is the image and S(d) is the structuring element of diameter d. Closing with a structuring element with a specific diameter leads to the filling of all (parts of) voids smaller than the size of the structuring element.
689
By choosing a set of increasing circular structuring elements, a cumulative size distribution G(d) can be obtained, as
GðdÞ ¼
∫C ðx; dÞdx−∫C ðx; 0Þdx ∫C ðx; ∞Þdx−∫C ðx; 0Þdx
;
ð2Þ
where the closing C(x,0) is equal to the original image and C(x,∞) is the closing with an infinite structuring element, which is equal to the image with all voids filled in. 3. Theory 3.1. Theory of ice crystal growth during freeze-drying The pores present in the freeze-dried vegetables are created via the formation of ice crystals during the freezing step and the subsequent sublimation of the ice crystals during the drying step. In most freezing operations of foods, ice crystals are formed via so-called dendritic growth (Chevalier, Le Bail, & Ghoul, 2000; Hartel, 1996; Woinet, Andrieu, Laurent, & Min, 1998). This form of crystal growth is investigated in great detail in the field of binary alloys, and in particular for the process of directional solidification (Hunt & Lu, 1996; Kurz & Fisher, 1981). During directional solidification one has observed that the crystalline dendrites have some regular interspacing that is defined by the freezing rate rather than the number of nuclei, as in the cryo-freezing of pharmaceuticals (Searles, Carpenter, & Randolph, 2001). This phenomenon is also assumed to occur in food freezing, although the freezing process of foods is not as well defined as in directional solidification. The difference between both processes is merely at the start of the crystallization. Food freezing will not start with a planar interface, but from several seeds formed by heterogeneous nucleation. For large sized foods the nucleation will start predominantly at the exterior surface of the food, which is exposed to the coolant. In fresh foods having retained a cellular structure at the start of freezing, the nucleation can be extracellular and/or intracellular. At slow freezing rates, the nucleation is extracellular, while at fast freezing rates as in cryogenic freezing the nucleation is mainly intracellular. The turgor pressure inside the cell makes it thermodynamically favorable that nucleation starts extracellular. It was shown (Mazur, 1984) that the change of intercellular ice nucleation depends on the ratio of freezing rate and the time scale of water permeation through the cell membrane. If the food is blanched prior to the freezing process, the cell membrane loses its integrity, and water can permeate freely, and consequently there is no preference for extracellular nucleation. Exact prediction of the dendrite interspacing is still a challenging problem. A variety of semi-empirical and theoretical scaling rules exists, but large efforts in numerical simulation of dendritic growth, have not yet led to a conclusive theory. A number of scaling rules for directional solidification that have appeared in literature are shown in Table 1. All these scaling rules for the dendrite interspacing, λ, are formulated in terms of the front velocity of the solidification front, R• , and the temperature gradient, G. The proportionality constant in the scaling rule has shown to be dependent on other conditions, like the initial concentration of the solute (Madaghiele, Sannino, Yannas, & Spector, 2008). As Table 1 Scaling rules for dendrite interspacing λ during solidification used in literature. Scaling rule λe R
−0:25 •
G
−0:5
•
λe R−0:5 G−0:5 •
λe R−1 G−1
Reference Kurz & Fisher, 1981; Butler, Reeds, & Dawson, 1981; Kochs, Korber, Nunner, & Heschel, 1991; Kurz & Fisher, 1981; Searles et al., 2001 Rohatgi & Adams, 1967; Bomben & King, 1982; Rohatgi, Jain, & Adams, 1968; Woinet et al., 1998 Tiller & Rutter, 1956; Bevilacqua et al., 1979
690
A. Voda et al. / Food Research International 49 (2012) 687–693
one can observe from Table 1 most of the (alloy) literature follows the scaling rule of Kurz and Fisher. It appears that the remainder of the scaling rules can be reformulated in terms of the freezing rate. It is assumed that the scaling rules for ice crystal size holds all freezing conditions, irrespective of whether the ice is nucleated intra- or extracellular (Bevilacqua, Zaritzky, & Calvelo, 1979). Hence, the integrity of the cellular tissue does not determine dendrite interspacing. Consequently, ice crystal formation in cellular food like vegetables can be compared with that in solutions and gels of biomaterials. Because in food freezing the front velocity and temperature gradient cannot be controlled independently as in directional solidification, we express the freezing conditions in terms of the freezing rate T • ¼ R• G. The freezing rate is estimated via the approximation of Planck, which takes the heat transfer coefficient of the coolant, h, and the coolant temperature, T0. Planck states that the freezing rate is equal to the difference of the initial freezing point of the food material and the coolant temperature, Tf − T0, divided by the freezing time tf, which is estimated as tf ¼
ΔH V ð1 þ Bi=4Þ: T f −T 0 hA
ð3Þ
Here ΔH is the latent heat of fusion for ice, V is the volume of the food sample, and A is its exterior surface area. Bi = hD/ks is the Biot number, with D the characteristic length scale of the food material (6 V / A), and ks the thermal conductivity of ice. 4. Results and discussion 4.1. Microstructure of freeze-dried carrots by SEM and μCT Information regarding the state of the dried microstructure and the integrity of the cellular tissue can be extracted from SEM images shown in Fig. 2a. The sublimation of the ice crystals grown within the carrot leaves a dried matrix representing a fingerprint of the ice crystals sizes and shapes. The samples frozen at lower temperature show smaller pores as the ice crystals are expected to grow less under fast cooling conditions. However, during freezing, the growth of an ice crystal ruptures, pushes, and compresses cells. This damage is more pronounced in slowly frozen tissue which yields bigger ice crystals. Another observation is that blanched samples show a finer microstructure. SEM images clearly reveal better microstructure preservation with increasing freezing rate in combination with blanching. The carrot sample
blanched and frozen at −196 °C consists of a cellular architecture that suffered less structural damage. On the basis of pore sizes of the samples frozen at the higher temperature we expect that the cell membranes can be damaged by the ice crystals formed. Microstructures of dried carrot samples determined by μCT are depicted in Fig. 2b. From a qualitative perspective, μCT images show the dry matrix consisting of pores and cavities left after sublimating the ice from the carrot tissue. The direction of the ice crystal formation can be clearly seen in the slow frozen samples. Slow freezing allows ice crystals to grow outside cells (Brown, 1976; Van Buggenhout et al., 2006), causing damage by cell collapse and rupture. Fast freezing determines ice crystals to grow inside cells with very little cell separation and much less damage. However, freezing via liquid nitrogen is prone to freeze cracking. In this case blanching before fast freezing was found to exhibit a beneficial effect and to reduce the structural damage. Therefore the impact of freezing rate on a non-blanched and blanched sample can be different, as it will be investigated later by NMR. Moreover, sampling the carrot pieces in cases when two (or more) distinct tissue zones are subject to thermal treatment and drying results in the appearance of randomly oriented mechanical stress and this causes additional microstructure damage. 4.2. Granulometry-based pore size distributions from μCT images of dried carrots The results of the granulometry computation for all datasets are shown in Fig. 3a. The granulometries of the samples with fine structures rise at small diameters, while the curves of samples with large voids rise at larger diameters. A characteristic value for every curve is derived as the diameter where the cumulative granulometric curve is 0.50, i.e. where half of the volume contains voids smaller than this characteristic diameter. The acquired eight characteristic diameters are shown in Fig. 3b. A standard deviation of the characteristic pore sizes can also be estimated from these granulometries. In the case of an ideal Gaussian distribution, 68% of the data points lie between ± one standard deviation of the mean. Although the granulometries are not ideal Gaussian distributions, we can still find this 68% region around the characteristic diameters and obtain a rough estimate of the standard deviations. By finding those diameters where the cumulative granulometric curves cross the values 0.16 and 0.84, we obtain the downside and upside positions of the 68% region. The acquired estimations for the standard deviation are not symmetric due to the fact that the curves themselves are not symmetric.
Fig. 2. (a) SEM and (b) μCT images of carrot tissue freeze-dried at −28 and −196 °C, with and without blanching pre-treatment.
A. Voda et al. / Food Research International 49 (2012) 687–693
691
Fig. 4. Ice crystal size against freezing rate, according to our experiment data (circles), and literature data (see supporting information). Empty circles are for non-blanched samples, and full ones are for blanched samples. Freezing rate is estimated via Planck's equation as discussed in the main text. The solid line is a regression of crystal size versus freezing rate: λ ¼ 50T • −0:25 , with crystal size λ in microns, and freezing rate T • in °C/s.
Fig. 3. (a) Granulometry of all eight methods (four different temperatures and two pretreatments), and (b) characteristic diameters of all eight methods (four different temperatures and two pretreatments) with their corresponding estimations of the standard deviation.
found in literature, as given in Table 1. However, the exponent in the power law is close to those given by Kurz and Fisher (Kurz & Fisher, 1981). Often, in frozen food there is a gradient in ice crystal size from the surface to the center (Bevilacqua et al., 1979), and this may well mean that the correlation of Kurz and Fisher holds for foods, as the crystal size gradient depends of course on the location.
4.3. Prediction of pore sizes in dried carrot To compare the average pore size determined by granulometry with the scaling laws for ice crystal growth and experimental data on ice crystal sizes for comparable (bio)materials, the applied freezing conditions in terms of heat transfer coefficient, h, and the coolant temperature T0 are needed (see supporting information). The heat transfer coefficient was estimated using values from literature (Heldman & Singh, 1981). With the given values of h and the values of D ≈10 mm and ks ≈ 2.7 W/m.K., a small Biot number is obtained, 0.05 b Bi b 1, implying that external limitations to heat transfer are dominant over internal limitations. Our data obtained for carrots will be compared to data from literature, describing ice crystal size as obtained by freezing. We have chosen (bio)materials having a comparable freezing point with vegetables, which also means similar initial concentration of solute which is just below the freezing point (as compared to the coolant temperatures). The data sources are listed in Table S2 in the supporting information. Self-diffusion of water in carbohydrate solutions and biopolymer solutions appears to show universal behavior—as we will show in a forthcoming paper. Hence, we may well compare the results of these materials. The pore sizes estimated from granulometry together with literature data (Bevilacqua et al., 1979; Ding, Huang, & Zhou, 1997; O'Brien, Harley, Yannas, & Gibson, 2004; Schoof, Bruns, Fischer, Heschel, & Rau, 2000; Ueno, Do, Sagara, Kudoh, & Higuchi, 2004) are shown in Fig. 4. It can be observed that all data collapse to a single curve, and our data fall on this common line—except for the blanched carrot frozen at − 28 °C. It was shown in a previous study that thermal pre-treatments have little impact on the morphology of frozen carrots (Van Buggenhout et al., 2006), but the respective pre-treatment occurred at 60 °C and not at the water boiling point as it was performed in this work. We have fitted the common line through the collection of data points (except for the outlier) to a power law, where we have obtained that λeT : −0:25 . This scaling is different from the ones
4.4. Morphology of rehydrated carrots by MRI and NMR relaxometry and diffusometry Water distribution in rehydrated (15 min) carrot samples was visualized by magnetic resonance imaging (Fig. 5a). Susceptibility artifacts can be observed due the air present in non-hydrated pores. The signal intensity in the images depends on the water amount and on the carrot tissue density per pixel. As a first observation, all samples show partial rehydration after 15 min, indicated by spots in the image with low signal intensity. A difference in the structure between the blanched and non-blanched slowly frozen samples can be identified; smaller non-hydrated cavities are present in the blanched samples. This is supported by visual inspection of corresponding SEM (data not shown) and μCT images (Fig. 2b). The dried microstructure revealed by μCT is reflected in with the water distribution in the rehydrated structure. Thereby, it is clear that pre-treatments have an impact on both the dried and rehydrated states. Furthermore, the water distribution in the rehydrated sample blanched and frozen at − 196 °C indicates microstructure details similar to a native carrot structure. Structural anisotropy was assessed by measuring water selfdiffusion coefficients along longitudinal and transversal directions (Fig. 5b, axial is defined as aligned with the static magnetic field and along the carrot root). The ratio between diffusivities along the two directions, estimated at long diffusion times (> 1 s), gives an indication about the anisotropic nature of the system. The anatomy of fresh carrots is known as bi-dimensional anisotropic, due to cells elongated along the direction of the growth. It was found out that anisotropy to a certain extent remains after thermal treatments and drying. Samples frozen at lower temperatures show more anisotropy, thus confirming that faster freezing preserves cell wall morphology. Micro-scale morphology in rehydrated freeze-dried specimens was assessed by means of transversal relaxation data on rehydrated
692
A. Voda et al. / Food Research International 49 (2012) 687–693
Fig. 5. Microscopic features of rehydrated cortical tissue of carrots freeze-dried at −28 and − 196 °C, with (B) and without (NB) blanching pre-treatment: (a) MRI, and (b) Water self-diffusion constants measured at Δ = 1 s along longitudinal and transversal directions.
carrots. Three relaxation components could be identified for all samples. The short (~20 ms) and intermediate (100–180 ms) T2's correspond to the nonexchangeable protons of the macromolecules (cellulose, carotenoids, sugars and pectin) and to the water present in the swollen cell walls. The main (long T2) component (400 up to 600 ms) corresponds to water filling the rehydrated pores, and it is shown in Fig. 6a for all carrot samples. It can be observed that the variation of relaxation times with the freezing temperature occurs on a relative narrow interval. Assuming the surface relativity of the solid carrot matrix is independent of the freezing rate, one would expect that the NMR relaxation rate would scale inversely proportional with the pore size, which is confirmed in the experimental data shown in Fig. 6a (see granulometry in Fig. 3). There are two factors reducing the relaxation time of the main T2 population in rehydrated carrots: the surface-to-volume ratio of the pore space and the chemical exchange between the water and sugar protons. The effect of proton exchange with sugars was investigated in model aqueous solutions and it revealed a reduction of T2 up to 1.5 s for a native sugar concentration in carrots (ranging from 5 to 20%), as compared to the T2 of pure water. Moreover, relaxation measurements were performed on water removed from a rehydrated carrot sample by
compression, and the T2 was found to be of the order of 1 s. The results presented in Fig. 6a indicate that transverse relaxation is further reduced by a significant amount, down to 500 ms, and we attribute this to pore surface relaxation. Hence differences in dry microstructures with characteristic pore sizes calculated by granulometry can be correlated with relaxation behavior of water in rehydrated state measured by T2-NMR. Cell wall network tortuosity was determined by fitting (Sen, 2004; Sibgatullin et al., 2010) the experimental self-diffusion data as a function of diffusion time (Fig. 6b). It can be observed that the tortuosity of slow frozen samples at −28 °C is very close to unity, which indicates that water molecules can diffuse through the porous structure with hardly any obstructions. This confirms that native (membrane separated) compartments have been destroyed after freeze-drying and that their functionality is not recovered after rehydration. The disruption of membranes makes the rehydrated system behave more like a sponge characterized by network tortuosity. The disruption of semi-permeable membranes and distinct compartments precludes the use of the numerical plant cell model (van der Weerd et al., 2002). The fast frozen samples show a significantly higher tortuosity (Fig. 6b). As it was already discussed (SEM, μCT), the pores are smaller in this case and their number per unit volume must be higher. For the water molecules to travel through the structure more connections are needed, assuming that the permeability of the rehydrated cell wall is independent of the applied cooling rate. Interestingly, the carrots blanched prior to freezing and drying show a higher tortuosity. Therefore, it appears that the intrinsic permeability of the pore walls of the rehydrated carrots is not dominant and it is not determining the tortuosity in this case. The SEM images of the fast frozen carrots shown in Fig. 2a indicate better microstructure preservation for the blanched carrots, with less damage on cell walls and less cell collapse, as discussed previously. The non-blanched carrot probably exhibits more connections between pores due to freezing damage at the cell level. Moreover, the difference in granulometries and average pore sizes is rather small (Fig. 3), which makes the non-blanched carrot look like a blanched one but with perforated cell walls. Hence, the network tortuosity of non-blanched systems reflects the connectivity of numerous pores and the permeability of the cell walls plays a role only if the surface to volume area varies, as it is the case of different pore sizes. On the other hand, during blanching membrane disruption will result in sugar homogenization over the tissue. Therefore, a blanched carrot can better accommodate intracellular ice crystal growth and will thus suffer less structural damage. This is not necessarily valid in the case of slow cooling where freezing occurs extracellularly and blanching may hardly have any effect. 5. Conclusions
Fig. 6. Micro- and macroscopic features of rehydrated cortical tissue of carrots freeze-dried at −196, −150, −80 and −28 °C, with and without blanching pre-treatment: (a) proton transverse relaxation time, T2, and (b): network tortuosity (from water self diffusion by NMR).
Cell wall network tortuosity and anisotropy of cortical tissue of winter carrots (diffusion NMR) indicate that fast freezing is beneficial to structure preservation. Slow freezing allows ice crystals to grow outside cells, causing damage by cell collapse and rupture. Fast freezing determines ice crystal growth inside cells with very little cell separation and much less damage. Blanching before freezing in general does not affect the pore sizes of the dry product. However, due to a more homogeneous distribution of sugar upon blanching, subsequent fast freezing will result in less damage to the cell wall architecture (SEM, diffusion and relaxation NMR). Under the assumption that pore sizes in the freeze-dried carrots are equal to ice crystal size, it was shown that our experimental data is in agreement with published data on other biomaterials with similar initial freezing points. All data for these materials fall on a single curve, which implies that the ice crystal size scales with the freezing rate by a power law, λ ~ T −0.25. This finding substantiates the fact that ice crystals in carrots have a dendritic growth, undergoing coarsening instability, rendering more or less regular interspacing between ice crystals.
A. Voda et al. / Food Research International 49 (2012) 687–693
Acknowledgment This work was supported by the Dutch Food and Nutrition Delta program (project FND080078U). Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.foodres.2012.08.019. References Bangham, J. A., Chardaire, P., Pye, C. J., & Ling, P. D. (1996). Multiscale nonlinear decomposition: The sieve decomposition theorem. IEEE Transactions on Pattern Analysis and Machine Intelligence, 18, 529–539. Bernstein, Matt A., King, Kevin F., & X. J. Z. (2004). Handbook of MRI pulse sequences. Bevilacqua, A., Zaritzky, N. E., & Calvelo, A. (1979). Histological measurements of ice in frozen beef. Journal of Food Technology, 14(3), 237–251. Bomben, J. L., & King, C. J. (1982). Heat and mass-transport in the freezing of apple tissue. Journal of Food Technology, 17(5), 615–632. Brown, M. S. (1976). Texture of frozen fruits and vegetables. Journal of Texture Studies, 7(4), 391–404. Butler, J. P., Reeds, J. A., & Dawson, S. V. (1981). Estimating solutions of 1st kind integral-equations with nonnegative constraints and optimal smoothing. Journal of Numerical Analysis, 18(3), 381–397. Chassagne-Berces, S., Poirier, C., Devaux, M. F., Fonseca, F., Lahaye, M., Pigorini, G., et al. (2009). Changes in texture, cellular structure and cell wall composition in apple tissue as a result of freezing. Food Research International, 42(7), 788–797. Chevalier, D., Le Bail, A., & Ghoul, M. (2000). Freezing and ice crystals formed in a cylindrical food model: part I. Freezing at atmospheric pressure. Journal of Food Engineering, 46(4), 277–285. Cornillon, P. (2000). Characterization of osmotic dehydrated apple by NMR and DSC. Lebensmittel-Wissenschaft Und-TechnologieFood Science and Technology, 33(4), 261–267. Devahastin, S., & Niamnuy, C. (2010). Modelling quality changes of fruits and vegetables during drying: A review. International Journal of Food Science and Technology, 45(9), 1755–1767. Ding, G., Huang, W., & Zhou, Y. (1997). History-dependence of dendritic tip radius and secondary arm spacing during unidirectional solidification. Journal of Materials Science Letters, 16(5), 376–378. Hartel, R. W. (1996). Ice crystallization during the manufacture of ice cream. Trends in Food Science & Technology, 7(10), 315–321. Heldman, D. R., & Singh, R. P. (1981). Food process engineering, 178–180. Hills, B. P., & Nott, K. P. (1999). NMR studies of water compartmentation in carrot parenchyma tissue during drying and freezing. Applied Magnetic Resonance, 17(4), 521–535. Hoffmann, K., Boeing, H., Volatier, J. L., & Becker, W. (2003). Evaluating the potential health gain of the World Health Organization's recommendation concerning vegetable and fruit consumption. Public Health Nutrition, 6(8), 765–772. Hunt, J. D., & Lu, S. Z. (1996). Numerical modeling of cellular dendritic array growth: Spacing and structure predictions. Metallurgical and Materials Transactions A, Physical Metallurgy and Materials Science, 27(3), 611–623. Jangam, S. V. (2011). An overview of recent developments and some R&D challenges related to drying of foods. Drying Technology, 29(12), 1343–1357. Kochs, M., Korber, C., Nunner, B., & Heschel, I. (1991). The influence of the freezing process on vapor transport during sublimation in vacuum-freeze-drying. International Journal of Heat and Mass Transfer, 34(9), 2395–2408. Kurz, W., & Fisher, D. J. (1981). Dendrite growth at the limit of stability—Tip radius and spacing. Acta Metallurgica, 29(1), 11–20. Luengo Hendriks, C. L., van Kempen, G. M. P., & van Vliet, L. J. (2007). Improving the accuracy of isotropic granulometries. Pattern Recognition Letters, 28(7), 865–872.
693
Madaghiele, M., Sannino, A., Yannas, I. V., & Spector, M. (2008). Collagen-based matrices with axially oriented pores. Journal of Biomedical Materials Research. Part A, 85A(3), 757–767. Mazur, P. (1984). Freezing of living cells—Mechanisms and implications. American Journal of Physiology, 247(3), C125–C142. Mebatsion, H. (2008). Modelling fruit (micro)structures, why and how? Trends in Food Science & Technology, 19(2), 59–66. Mujumdar, A. S. (2011). Editorial: Development of new industrial drying technologies. Drying Technology, 29(11), 1249–1250. Mujumdar, A. S., & Law, C. L. (2010). Drying technology: Trends and applications in postharvest processing. Food and Bioprocess Technology, 3(6), 843–852. O'Brien, F. J., Harley, B. A., Yannas, I. V., & Gibson, L. (2004). Influence of freezing rate on pore structure in freeze-dried collagen-GAG scaffolds. Biomaterials, 25(6), 1077–1086. Pomerleau, J., Lock, K., & Mckee, M. (2006). The burden of cardiovascular disease and cancer attributable to low fruit and vegetable intake in the European Union: Differences between old and new member states. Public Health Nutrition, 9(5), 575–583. Prothon, F., Ahrne, L., & Sjoholm, I. (2003). Mechanisms and prevention of plant tissue collapse during dehydration: A critical review. Critical Reviews in Food Science and Nutrition, 43(4), 447–479. Provencher, S. W. (1982). Contin—A general-purpose constrained regularization program for inverting noisy linear algebraic and integral-equations. Computer Physics Communications, 27(3), 229–242. Ratti, C. (2001). Hot air and freeze-drying of high-value foods: a review. Journal of Food Engineering, 49(4), 311–319. Rohatgi, P. K., & Adams, C. M., Jr. (1967). Ice-brine dendritic aggregate formed on freezing of aqueous solutions. Journal of Glaciology, 6, 47,663–47,679. Rohatgi, P. K., Jain, S. M., & Adams, C. M., Jr. (1968). Dendritic crystallization of ice from aqueous solutions. Industrial and Engineering Chemistry Fundamentals, 7(1), 72–79. Sagar, V. R., & Kumar, P. S. (2010). Recent advances in drying and dehydration of fruits and vegetables: A review. Journal of Food Science and Technology-Mysore, 47(1), 15–26. Schoof, H., Bruns, L., Fischer, A., Heschel, I., & Rau, G. (2000). Dendritic ice morphology in unidirectionally solidified collagen suspensions. Journal of Crystal Growth, 209(1), 122–129. Searles, J. A., Carpenter, J. F., & Randolph, T. W. (2001). The ice nucleation temperature determines the primary drying rate of lyophilization for samples frozen on a temperature-controlled shelf. Journal of Pharmaceutical Sciences, 90(7), 860–871. Sen, P. N. (2004). Time-dependent diffusion coefficient as a probe of geometry. Concepts in Magnetic Resonance Part A, 23A(1), 1–21. Sibgatullin, T. A., Vergeldt, F. J., Gerkema, E., & Van As, H. (2010). Quantitative permeability imaging of plant tissues. European Biophysics Journal with Biophysics Letters, 39(4), 699–710. Tiller, W. A., & Rutter, J. W. (1956). The effect of growth conditions upon the solidification of a binary alloy. Canadian Journal of Physics, 34, 96–121. Ueno, S., Do, G. S., Sagara, Y., Kudoh, K., & Higuchi, T. (2004). Three-dimensional measurement of ice crystals in frozen dilute solution. International Journal of Refrigeration— Revue Internationale du Froid, 27(3), 302–308. Van Buggenhout, S., Lille, M., Messagie, I., Van Loey, A., Autio, K., & Hendrickx, M. (2006). Impact of pretreatment and freezing conditions on the microstructure of frozen carrots: Quantification and relation to texture loss. European Food Research and Technology, 222(5–6), 543–553. van Dalen, G., Notenboom, P., van Vliet, L. J., Voortman, L., & Esveld, E. (2007). 3-D imaging, analysis and modelling of porous cereal products using X-Ray microtomography. Image Analysis Stereology, 26, 169–177. van der Weerd, L., Melnikov, S. M., Vergeldt, F. J., Novikov, E. G., & Van As, H. (2002). Modelling of self-diffusion and relaxation time NMR in multicompartment systems with cylindrical geometry. Journal of Magnetic Resonance, 156(2), 213–221. van Duynhoven, J., Voda, A., Witek, M., & As, Van (2010). Time-domain NMR applied to food products (pp. 145–197). Woinet, B., Andrieu, J., Laurent, M., & Min, S. G. (1998). Experimental and theoretical study of model food freezing. Part II. Characterization and modelling of the ice crystal size. Journal of Food Engineering, 35(4), 395–407.